Right _and_ wrong, your diagnostic videos are just the best. You explain your logic, why you move from component to component, as well as referencing the diagrams/docs and you show us ever bit of it. I really appreciate these videos. Thanks.
I think you should do a tour of your workbench area. It looks very cozy, and very well laid out with anything you need at any moment in arm's reach. But we don't know what everything is, or what things you have that are there but not apparent, or why you have certain things, or why the placement matters. It would probably help others (like me!) in laying out their own workbenches for working on not only vintage computers, but other electronics. Thanks for all the videos!
Adrian you are a true expert as you refer to yourself as only a novice. A true expert knows there is always something new to learn. Someone could have extensive knowledge on this particular motherboard and repair it quicker but the real talent is in looking at something new to you and being able to troubleshoot and fix and make sense of it as you do. Every time you go over traces you are examining someone’s work from about 40 years ago which is very cool! It is amazing these devices can continue to work perfectly so many years later. Great work!
U84 can be fitted with a quad 2 input multiplexer, like a 74ls158. You can then if i'm not mistaken use 256x1kb chips in all memory Banks. Replacing U44 with a GAL with proper equations might even allow you to have as much as full 1MB and map RAM on UMB areas not used by hardware. Seems that IBM was thinking ahead.
Yeah, I did a similar thing on my overclocked Compaq Portable board (the quintessential IBM 5150 clone). This thing, like the 5150 only has 5 expansion slots and 256k on board so maxing out the on board memory (to 640k in my case) comes in really handy as all my slots are populated with cards. The 5160 has 8 slots so it's less likely to become overpopulated by expansion cards, leaving enough free slots for a memory expansion card in most configurations. It would a nice gimmick though to have a 5160 with a full 1mb (dude that's gnarly!) on board instead of that measly 64-256k :-)
It may also be that they were actually thinking _back._ While IBM had never done anything else that was compatible with the IBM PC, they _had_ made somewhat similar computers before (the IBM Datasette series I think?), so they certainly had experience with iterating on 8088/8086 designs.
@@absalomdraconis They actually contemplated the 6502, 8085, Z80, and 68000, as well as their very own CPU design for the IBM PC project. But for some reason, they selected the 8088. Perhaps because what you wrote and that they also used the slightly older 8085 and its support ICs. As you may know, several 8080/85 family ICs were used in the IBM PC. Pretty compatible with the 8088 (largely also with the 8086). Not sure it was the same designers though, as they had multiple teams.
Kudos for the patience for this... 1/3 of the way through this video, I was thinking I'd already have cut out that 245 and soldered in a socket.. I'm not used to seeing floating busses like that either so the scope signals definitely drove me in that direction.. I would REALLY love to see you read that dead ROM. it would be interesting to see what it has. Is one bit stuck on all bytes? Is it just random data? are just a few bytes wrong, or does it completely fail and read all 0 or ff? inquiring minds want to know!!!!!
Just noticed an interesting detail: Remember how Adrian initially recieved the "stack failure" message when trying to access the floppy drive? This must have been due to the same floppy reading routine failue (caused by the failed ROM chip) he expericenced with the diagnostic ROM installed. So the IBM ROM chip must have died in the time between Adrian accessing the drive with the cleaning disk inserted and him accessing it with the DOS 3.3 disk inserted. Or possibly even before that, as the drive obviously couldn't deliver valid data when trying to read the cleaning disk. I certainly wonder why and how the chip failed... And of course great and especially well explained diagnostics, Adrian! Huge thanks for sharing the whole process, including when you went down that "rabbit hole". But just as a reference: I'm currently trying to repair a modern motherboard, and it's certainly possible that i've almost only gone down rabbit holes so far... Repairing really is a matter of experience, and thus i'm really grateful for being able to learn from you. Keep up the good content!
I really don't understand much of what you were doing but I worked doing electrical diagnosis for a large Toyota dealership and did work with basic test equipment and sometimes with an old Tectronic analog scope and later a digital scope to analyze wonky signals from various EFI sensors. I also built early computers from parts starting with 286's running DOS and later 386's, 486's etc. So I know what the inside of a PC looks like and have a rudimentary understanding of the various components inside one. I just like watching people diagnosing any type of electronic equipment. I liked the way you explained your process as you went along. I imagine it was a game changer to be able to program and use an EPROM to stand in for various faulty chips. I was watching a group of techs try to bring to life an actual Apollo guidance computer and they had some young kid who was able to program FPGA's to stand in for faulty and otherwise unobtainable hardware modules. Anyway the bottom line is the Mounties got their man and found the problem. Will be watching and learning from more of your videos in the future.
Another awesome video Adrian, Love your thought process and diagnostics checking! but you always manage to get a positive result in the end and it must feel so rewarding! Keep all your brilliant video's coming please.
I love your videos, not only because they're interesting, but because you leave in EVERYTHING, warts and all. It's nice to watch the process, even when you go off in the wrong direction. This is a fantastic channel!
Did anyone else get Eye of the Tiger to start up in their mind when he called that beautiful Motherboard a survivor? Great video! I'm so glad he could fix it!
The C, D or P in front of a device number on an Intel device denotes the package type. C will be standard ceramic, D will be dual layer ceramic and P will be epoxy. It's interesting that AMD used Intel's package type notation as well.
@Lassi Kinnunen 81 That is basically correct. Intel actually spun off AMD as a separate manufacturer of intel's chips to help with batch loads. That way Intel could keep up. Eventually AMD came up with their own designs and had to go their own way to make them since Intel didn't want the competition and tried to keep them for making their own chips. It's actually a lot more complicated than that but that's the general story.
I enjoy your videos immensely, even when you go in the wrong direction it's worth every second. Keeping old tech history alive and learning new techniques along the way, thanks Adrian.
It is interesting to watch your thought process in trouble shooting, and that alone is something people can learn from. But for me, having worked on these old PCXT systems, I was struggling watching you bark up all the wrong trees but I honestly share your excitement when you figure it out. This was a lot of fun!
The honesty in the process, mistakes, etc should be reassuring to everyone that runs into problems. Also how critical thinking can take you down all kinds of paths and ultimately down the right one if you're patient enough and persistent. I work more on clockwork and electromechanical projects than electronics. It's fun to watch videos like these to go back to when I was doing electronic engineering and explore the problem solving process along with.
This is one of my favourite ADB videos yet, Adrian. The reason is because you go through the diagnostic process in a very natural way. It also covers more familiar territory for me: IBM stuff
A very good example of fault finding. You call yourself a novice, but i think you undersell yourself, which is a humble character trait and certainly no bad thing. Keep the great memory lane videos coming!
I seem to remember you repairing a C64 and it was the kernal rom that was taking the whole system down (even thought the kernal wasn't required to run the diagnostic). So i think you've been down this rabbit hole before.
Going down the wrong path is frustrating but often illuminating and you rarely get tripped up the same way again, well done and thanks for sharing this oft overlooked part of the repair process!
I was testing an Altos CPM computer in the 80's which had ceramic ram chips and found the faulty chip when I left my finger print behind on it as it must have been so incredibly hot. At the time I thought my skin would not grow back normally but after a couple of years I was back to normal. From that point on I put a sheet of paper on the chips first to feel if they where cool enough to touch with bear fingers so watch out! It always was a good diagnostic test. Also reseating the chips in sockets.
I made the same mistake troubleshooting a C64. All the diagnostics pointed to bad ram...ended up being a bad Kernel ROM. Spun my wheels a lot that day. Congrats on getting that board working!
@Adrian You're better a bench-tech than you think you are. Novice you are not. BIOS chip failures are very rare and can often lead to odd diagnosis results like what you experienced. Trust me on this poing, compared to most people who work on computers, you have a high level of expertise. But you on the humble side of things and clearly have a desire to continue learning. Hang on to that desire. It will continue to serve you well!
This channel is so unique to me as i don't understand anything yet sit and watch the videos from start to finish. I find myself wanting to buy an oscilloscope and look at waveforms!! Thank you for the content
"I'm still quite the novice". Oh, come on, being humble honors you, and I appreciate it. But it is not exactly like many true experts in repairing those very early PCs in this time and age exist anymore. At present, for nearly everyone, it's about finding your way through it, and we love watching the problem solving process, and the convoluted way that leads to an exciting FRIGGIN' WORKS finale! You're the hero of our favourite tech adventures, keep up the great work.
Adrian Don't forget to cover the window over the 27??? EPROM, as I was using them in the early days and if they were not covered we found that strong uv light would currupt the data. We used to put a rectangular one on with programm title issue and number of EPROM. I think the max we used was somthing like 12, but it was a number of years ago app 40. It think it was when NASA was taking nearly all the manufacturer of special 74 series chips for the Saturn moon shot. I think it was the 7414 schmitt-trigger ones and others. The early IBM XT and AT models were brilliant our problem at Vauxhall (GMC) in Luton UK was sending the larg amount of data on the engine tests, back to Flint USA it was a nightmare, the systems would go down, regulary. It would have been quicker to fly the data there. Component level fault finding was my fortay, even used hot 🔥 fire lighters to diagnose faults on Ferranti computer boards the ceramic chips would work up to fantastic temperatures. Or freeze them. Thanks for another informative 👍 technical video.
For your virtual bench, maybe try something like putting a 100 or 10 ohm in series with the ground for the scope probe? I wonder if you're getting AC transients across whatever isolation it has that's glitching into the digital side as a ground bounce. Alternatively try using a short and robust USB cable so the ground can't bounce as easily. The resistance of the GND leads to the PC basically determines how far ground could drop based on the transient. Another option is to get a proper USB isolator that will get rid of the current path even more effectively if that's what it's doing.
Just found out about your channel and went down a deep dive into your videos! Love them! I wince though, every time you handle motherboards/cards/chips without a ground strap! Your basement temps is around 25C-26C based on some vids of yours and that would likely mean your relative humidity will be very low in a basement in Oregon. Low RH means you are likely a walking static bomb, frying cmos/mos parts as you go! Please check your RH and meybe think of a ground strap for one of your wrists. Love the channel, love your work, subscribed and sharing it!
You said it failed while you were using it, maybe it was a seating issue that showed up after the components warmed up and expanded. Maybe even dust in the socket? It would be fun, as others have said, to see what that rom Holtz. Maybe it is something that will work for other boards, even though EPROMs are a dime a dozen. It may in fact be un-x-orable. I always dig your videos! Keep up the great work.
When I worked at Gateway 2000 the 486DXLB motherboards were coming back bricked (no post). My coworker and I thought we would try something radical. We pulled the BIOS off of other bad boards of the same model and plugged them in the bricked ones. It fixed them every time. Oddly, service did not have a way to burn blank bios chips.
I love seeing the full experience incl. the false paths (something Usagi Electric also does a lot). I have definitely been there and given up too early. However, I can see that I probably need to get an (E)EPROM reader/programmer as that's very likely an issue I'm facing with my 1981 Nascom 2.
1x vote for a video attempting to fix USB issue on your modern motherboard!! Also if it is just front end ports check the connector/header, they and be annoyingly fragile. Awesome content as always! :)
Excellent job troubleshooting! Had the same issue with my Apple IIe. Ended up swapping the RAM only to realize that the issue wasn't the memory but rather one of the ROMs having failed *facepalm*
Adrian - i love this! i am a IT tech for last 26 years and this stuff was before my time but i love it. this video makes me feel stupid coz you are so smarter than me! but its awesome. and we all make mistakes, it is cool to watch this stuff! Keep it up!
It's good to learn and breakthrough with new way of thinking, learn from mistake feel good. Well, the sticker on the ROM is better for identification more than preventing erase, since I can't program the 2716 with the TL866II, I've switched to use EEPROM equivalent (I made adapter for 27c512 to 2716/32) , for faster erase and programming, the ultra violet thing waste lot of time, I keep some EPROM for collection but try not to use them.
I worked on server boards for almost 10 years for NCR. You are working blind without logic analyzers, digital scopes, and test PROMS. And by the time the boards got to me they had passed basic tests on the Fairchild 333 bed of nails fault finder. All failures went to a specific address I could trigger on and trigger out to the digital scope to see what was going on in real time on the board.
Hi, amazing work! Could you please put these videos in playlists? I cannot afford following the videos immediatelly as they're released and I am losing context if I missed something from the subject and looking for previous videos might be a struggle. You're meaning that MB blackscreened in the past video, now I am wondering which one... I love your content and would like to keep up with your projects :)
When I was restoring a certain 386 motherboard I decided to populate the board with filtering capacitors where the manufacturer cheaped out on them. All the spots where the power lines, except a single spot where it was 2.5V. Turned out it was IRQ2 line and it was pulled with 10k to GND and +5V at the same time.
Maybe start with the diagnostic ROM in the first place. That helps you eliminate bigger swaths of potential faults from the get go. Good job fixing it.
I keep meaning to try figuring out what's wrong with my 5170 motherboard. It's actually got a similar issue to your 5160 board, it doesn't do anything when powered on. Now that I know that minuszerodegrees has diagnostic ROMs for these boards, I should try that out. A POST card would probably come in handy too, although I feel like it's not even getting to that point.
Regarding PALs, there are (expensive) devices to read working chips, save the contents, and write new ones. I observed a fellow with such a machine working on old classic coin op arcade games and he replaced a few in some of the videos I watched. It's rather esoteric compared to burning an EEPROM.
Floating, un-terminated data/address lines can cause some interesting, intermittent, hard to diagnose failures. Many years ago I remember a particular piece of education software for the old Apple that would intermittently crash. Remember the Franklin/Apple knockoff computer? Well that same piece of education software would not run at all on the Franklin. It seems the Franklin motherboard was a bit larger than the Apple. It looked like Franklin just simply optically increased the physical size of the board but kept the IC socket sizes locations and everything else the same. I can't remember if it was a data or an address line but it ran all the way around the periphery of the motherboard. That little extra trace length on the Franklin motherboard, coupled with the floating data/address line, created a timing issue that prevented the program from running on the Franklin. Installing 2k pull-up or pull-down, (can't remember which it was) resistors on these all of these lines solved the issue for both the Apple and Franklin. This same program also had an issue with inconsistent and poor video colors caused by a missing pull-up resistor and a cap in the video modulator. (This was much worse on the Franklin than the Apple.) An added cap and a 2.2k pull-up resistor took care of this issue as well. I spent a few days adding these mods to all of the Apple II and Franklin computers used by a local, large public school system. This was a few years before they got an IT department.
If I had to guess about that empty socket, and it is completely a guess because I frankly have never even seen one of these motherboards, I’d guess part of a test harness or something used to test the board in production. The pins on the bottom side would make an ideal spot to contact on a bed of nails fixture, it it could have some kind of ROM chip in it to run test code.
I really appreciate you showing the IC was backwards; makes me feel much better about the socket I put in backwards while putting the Harlequin ZX Spectrum together :-)
Hi Adrian. Really enjoyed this video. It was great. Thanks for sharing your troubleshooting process. Please keep making this type of video so we can learn with you.
For your test bench rebooting, try putting it on its own UPS to help isolate it. If that works, there are cheaper fixes than dedicating a UPS permanently, but it's an easy enough way to find out. I have a monitor that likes to power cycle every time the air conditioning switches on, so I put a power conditioner on the line, and even that only mostly (~80%) fixes the problem.
Side thought based on your comment about your lab PC having dead USB ports -- if you have dead USB ports, 'tis the season for monitor flickers and dead USB ports due to static electricity in the winter. One fix I do that works pretty reliably is to pop out the BIOS battery and make sure all the power has drained from the mobo and give it about 15 minutes, and that often resets dead USB ports. Maybe give that a shot?
It's good you have electrical engineering education and experience of some form I have been doing ee as a hobby for a while now I find myself fixing Pentium iii and IV or even earlier motherboards the newer Post lga 775 boards have their idiosyncrasies I was given one with a celeron d allegedly a PC tech said it was unrepairable well just as you do I fixed it but it was a different issue but good work mate keep the videos coming I always learning from your videos
Chips can be the most resilient piece of hardware or can fail just by touching them if you have some static on you. On that regard, I learned that the hard way with a bios on an 3d gen Intel Core motherboard about 5 years ago, and that hurt because it was my main PC. Fortunately I already had the same IC programmer as you and I ordered the same bios chip so I could repair it, but also ordered an anti static strap and I always work with it every time now.
It's fascinating to see how the diagnostic chip reported the RAM IC being reversed. I did think that was a sneaky move to prove how it can detect these changes but probably risky if it did kill it. As it turns out, they are quite resilient to this sort of thing. I'm surprised that the board didn't beep with the CGA display being absent though. I suppose this was introduced a little later on.
my 5160 (640k board with a MDA) beeps when the video memory check fails. perhaps it was indeed not present on older boards, or adrian's board is jumpered incorrectly (unlikely, because CGA works once plugged).
Adrian, if you haven't already found it, and if my info is of any use, my IBM Personal Computer XT Technical Reference (6936808), Appendix D, Logic Diagrams, page D-8, System Board (Sheet 6 of 10) shows U84 as being an LS158, a Quadruple 2-Line to 1-Line Data Selector/Multiplexor (which I had to look up). But it does show pins 1, 2, 3, 4, & 15 (GND) as being connected, and of course, it doesn't even bother showing pin 16 (VCC). Since it shows more pins connected than your board appears to have connected, I guess my logic diagram is for a different version of the System Board?
Adrian I'm not sure whether or not you are using the TL866 or TL866 II programmer but the software and firmware has had some pretty major updates, I just updated to version 11 and the interface has changed a bit. The logic chip database is now much larger, the programmer will now attempt to automatically identify chips (it gets it right about 90% of the time), and you can now input your own test parameters for identifying chips. Its pretty great
lol adrian the mobo on your thumbnail blends in and looks like a mobo christmas sweater also the expression on your face makes me think of doc brown GREAT SCOTT haha
We’ve all been led down the wrong path when trying to chase down a fault in a circuit board, Adrian. It can be a mix of a DOH 🤦♂️ moment followed by a “it freaking works” moment when you find the actual problem and fix it…
I had to diagnose RAM issues on a Hyundai XT clone I bought off eBay. No technical reference manuals available, so I ended up reverse engineering a good chunk of the board to point me in the right direction (based on the symptoms I was getting). Turned out to be a faulty 74x00, one of the NAND gates had gone bad and it couldn't select the active bank properly. These are my favourite class of systems to diagnose and repair. IBM documented the truth tables for the PLAs, so one of those could be replaced if you needed. The issue, as you said, is with some of the clones like Tandy, Hyundai etc, where they'd integrated a bunch of the logic into ASICs. In which case, if you can't find a donner machine, you're probably hooped.
The osc software, the reason it faults while cycling devices on AC is because there's a sudden power spike, which can cause the power supply protection to shut off the device and keep it off till it's removed from power There's some ways of avoiding this like an isolated powerpoint for device's your testing, a good spike suppression plug and/or power board, or your best option is a EMI/EMC/RFI inline filter
I was gonna say (but you did cover it in the end), you filmed the death of that ROM. Watching the bus, you can see the ROM program running, then quit and halt the machine. The bus noise was just a distraction. It does make me wonder if adding a pull-up resistor to these machines would make them run more stable.
Would make them more reliable, as TTL pull up is pretty weak compared to the pull down, though better would be to add mid level terminators instead, either the 330R 470R series type, or the later integrated weak buffer CMOS type that actively held the line at 0 or 1 till a gate drove the weak buffer to the opposite state.
With LS TTL designs, open circuit buses are best left unless the design requires specific levels when the bus is not actively driven. If you pull signal lines high, you create more system noise when the lines are driven low. And mid-point termination is seldom needed. It's main purpose would normally be impedance matching, but at the edge speeds in a 74LS system with CPUs below 10MHz, it's not an issue. IBM did add terminators for the DRAM array where the distributed bus capacitance on address and control lines makes it act more like a transmission line, and that would have been normal then, just as it is today. Back in the mid-80's I did once have to put a pull-up on a memory bus on a card I had designed - the 16 floating data lines were being seen by the parity chips which were feeding back noise into the memory array and the whole thing oscillated at about 100MHz... A weak pullup (4.7k I think) fixed it.
I read the XT BIOS listings very thoroughly in the old days when I still had that kind of patience and focus, and I remember thinking the stack error would probably never happen because if there was something wrong that would cause it, it would almost 100% likely also cause either problems before that. So that was you first red herring right there.
I'm sure I mirror alot of people here when I say this, even the most experienced have moments like this. It's easy to get dragged down the wrong path, what's important out of all of that is that you did not give up and found the answer. That right there is what separates good tech troubleshooters versus the ones that are bad :)
Fantastic video. I always enjoy watching peoples diagnostic procedures. Can I ask - the eeprom programmer you have - is it a real mini pro? Or a TL866II clone? Are you able to post a link of where you purchased it from? Many thanks and keep up the great work :-)
Very funny, I made the same mistake troubleshooting my Commodore PC 10. Swapped the RAMs and the 245 but still the same. In the End there was no fault, was a bad Program Version that caused Freezes. And the half height Levels we're normal. Spended two days fixing a perfectly working machine. 🤣
I was an IBM service technician.
This brings back so many memories.
Thanks for the nostalgic video.
nostalgic is an understatement :)
"so many memories"
i see what you did there
What do you do now?retired?
If you're a novice, Adrian, then many of us are toddlers. I really appreciate your methodic approach to diagnostics and explanations. Thanks!
Zygote level myself.
If he's a novice, I'm an amoeba lol
I'm a future sperm.
Adrian is one of the finest technicians this old technician has had a priviledge to witness.
Nice work!
Right _and_ wrong, your diagnostic videos are just the best. You explain your logic, why you move from component to component, as well as referencing the diagrams/docs and you show us ever bit of it. I really appreciate these videos. Thanks.
I think you should do a tour of your workbench area. It looks very cozy, and very well laid out with anything you need at any moment in arm's reach. But we don't know what everything is, or what things you have that are there but not apparent, or why you have certain things, or why the placement matters. It would probably help others (like me!) in laying out their own workbenches for working on not only vintage computers, but other electronics. Thanks for all the videos!
Adrian you are a true expert as you refer to yourself as only a novice. A true expert knows there is always something new to learn. Someone could have extensive knowledge on this particular motherboard and repair it quicker but the real talent is in looking at something new to you and being able to troubleshoot and fix and make sense of it as you do. Every time you go over traces you are examining someone’s work from about 40 years ago which is very cool! It is amazing these devices can continue to work perfectly so many years later. Great work!
I like Adrian's joy when he solves a tough problem.
U84 can be fitted with a quad 2 input multiplexer, like a 74ls158. You can then if i'm not mistaken use 256x1kb chips in all memory Banks. Replacing U44 with a GAL with proper equations might even allow you to have as much as full 1MB and map RAM on UMB areas not used by hardware.
Seems that IBM was thinking ahead.
Yeah, I did a similar thing on my overclocked Compaq Portable board (the quintessential IBM 5150 clone). This thing, like the 5150 only has 5 expansion slots and 256k on board so maxing out the on board memory (to 640k in my case) comes in really handy as all my slots are populated with cards.
The 5160 has 8 slots so it's less likely to become overpopulated by expansion cards, leaving enough free slots for a memory expansion card in most configurations. It would a nice gimmick though to have a 5160 with a full 1mb (dude that's gnarly!) on board instead of that measly 64-256k :-)
@@pipschannel1222 Fun fact: In times of core memory the rule-of-thumb for good memory prices was "a dollar per bit".
IBM was always thinking ahead, at least back then.
It may also be that they were actually thinking _back._ While IBM had never done anything else that was compatible with the IBM PC, they _had_ made somewhat similar computers before (the IBM Datasette series I think?), so they certainly had experience with iterating on 8088/8086 designs.
@@absalomdraconis They actually contemplated the 6502, 8085, Z80, and 68000, as well as their very own CPU design for the IBM PC project. But for some reason, they selected the 8088. Perhaps because what you wrote and that they also used the slightly older 8085 and its support ICs. As you may know, several 8080/85 family ICs were used in the IBM PC. Pretty compatible with the 8088 (largely also with the 8086). Not sure it was the same designers though, as they had multiple teams.
I appreciate you showing the trouble shooting parts and not editing it out just to look better. Helps to show thought process involved in diagnosis.
Most of us that do electronics repair have gone down the rabbit hole many times. Good video.
Kudos for the patience for this... 1/3 of the way through this video, I was thinking I'd already have cut out that 245 and soldered in a socket.. I'm not used to seeing floating busses like that either so the scope signals definitely drove me in that direction.. I would REALLY love to see you read that dead ROM. it would be interesting to see what it has. Is one bit stuck on all bytes? Is it just random data? are just a few bytes wrong, or does it completely fail and read all 0 or ff? inquiring minds want to know!!!!!
Just noticed an interesting detail: Remember how Adrian initially recieved the "stack failure" message when trying to access the floppy drive? This must have been due to the same floppy reading routine failue (caused by the failed ROM chip) he expericenced with the diagnostic ROM installed. So the IBM ROM chip must have died in the time between Adrian accessing the drive with the cleaning disk inserted and him accessing it with the DOS 3.3 disk inserted. Or possibly even before that, as the drive obviously couldn't deliver valid data when trying to read the cleaning disk. I certainly wonder why and how the chip failed...
And of course great and especially well explained diagnostics, Adrian! Huge thanks for sharing the whole process, including when you went down that "rabbit hole". But just as a reference: I'm currently trying to repair a modern motherboard, and it's certainly possible that i've almost only gone down rabbit holes so far... Repairing really is a matter of experience, and thus i'm really grateful for being able to learn from you. Keep up the good content!
I really don't understand much of what you were doing but I worked doing electrical diagnosis for a large Toyota dealership and did work with basic test equipment and sometimes with an old Tectronic analog scope and later a digital scope to analyze wonky signals from various EFI sensors. I also built early computers from parts starting with 286's running DOS and later 386's, 486's etc. So I know what the inside of a PC looks like and have a rudimentary understanding of the various components inside one. I just like watching people diagnosing any type of electronic equipment. I liked the way you explained your process as you went along. I imagine it was a game changer to be able to program and use an EPROM to stand in for various faulty chips. I was watching a group of techs try to bring to life an actual Apollo guidance computer and they had some young kid who was able to program FPGA's to stand in for faulty and otherwise unobtainable hardware modules. Anyway the bottom line is the Mounties got their man and found the problem. Will be watching and learning from more of your videos in the future.
Another awesome video Adrian, Love your thought process and diagnostics checking! but you always manage to get a positive result in the end and it must feel so rewarding! Keep all your brilliant video's coming please.
Nice shirt !
I love your videos, not only because they're interesting, but because you leave in EVERYTHING, warts and all. It's nice to watch the process, even when you go off in the wrong direction. This is a fantastic channel!
Did anyone else get Eye of the Tiger to start up in their mind when he called that beautiful Motherboard a survivor? Great video! I'm so glad he could fix it!
The C, D or P in front of a device number on an Intel device denotes the package type. C will be standard ceramic, D will be dual layer ceramic and P will be epoxy. It's interesting that AMD used Intel's package type notation as well.
@Lassi Kinnunen 81 That is basically correct. Intel actually spun off AMD as a separate manufacturer of intel's chips to help with batch loads. That way Intel could keep up. Eventually AMD came up with their own designs and had to go their own way to make them since Intel didn't want the competition and tried to keep them for making their own chips. It's actually a lot more complicated than that but that's the general story.
I enjoy your videos immensely, even when you go in the wrong direction it's worth every second. Keeping old tech history alive and learning new techniques along the way, thanks Adrian.
It is interesting to watch your thought process in trouble shooting, and that alone is something people can learn from. But for me, having worked on these old PCXT systems, I was struggling watching you bark up all the wrong trees but I honestly share your excitement when you figure it out. This was a lot of fun!
As they say, it's not the destination that matters, but the journey.
The honesty in the process, mistakes, etc should be reassuring to everyone that runs into problems. Also how critical thinking can take you down all kinds of paths and ultimately down the right one if you're patient enough and persistent.
I work more on clockwork and electromechanical projects than electronics. It's fun to watch videos like these to go back to when I was doing electronic engineering and explore the problem solving process along with.
I remember working on those back in the late 80s and early 90s. Thanks for the trip down memory lane.
This is one of my favourite ADB videos yet, Adrian. The reason is because you go through the diagnostic process in a very natural way. It also covers more familiar territory for me: IBM stuff
A very good example of fault finding. You call yourself a novice, but i think you undersell yourself, which is a humble character trait and certainly no bad thing. Keep the great memory lane videos coming!
I seem to remember you repairing a C64 and it was the kernal rom that was taking the whole system down (even thought the kernal wasn't required to run the diagnostic). So i think you've been down this rabbit hole before.
Definitely although those MOS ROM failures are kind of expected. Normally, mask ROMs don't seem to fail.
Going down the wrong path is frustrating but often illuminating and you rarely get tripped up the same way again, well done and thanks for sharing this oft overlooked part of the repair process!
I was testing an Altos CPM computer in the 80's which had ceramic ram chips and found the faulty chip when I left my finger print behind on it as it must have been so incredibly hot. At the time I thought my skin would not grow back normally but after a couple of years I was back to normal. From that point on I put a sheet of paper on the chips first to feel if they where cool enough to touch with bear fingers so watch out! It always was a good diagnostic test. Also reseating the chips in sockets.
lol, you're excitement when something works never gets old Adrian!
I made the same mistake troubleshooting a C64. All the diagnostics pointed to bad ram...ended up being a bad Kernel ROM. Spun my wheels a lot that day. Congrats on getting that board working!
@Adrian
You're better a bench-tech than you think you are. Novice you are not. BIOS chip failures are very rare and can often lead to odd diagnosis results like what you experienced. Trust me on this poing, compared to most people who work on computers, you have a high level of expertise. But you on the humble side of things and clearly have a desire to continue learning. Hang on to that desire. It will continue to serve you well!
This channel is so unique to me as i don't understand anything yet sit and watch the videos from start to finish. I find myself wanting to buy an oscilloscope and look at waveforms!!
Thank you for the content
I would try to read that damaged ROM using the programmer (if possible) - to see if just some contents are damaged, or what.. :D
Wow! What a diag session, a marathon that ended up teaching so much.
"I'm still quite the novice". Oh, come on, being humble honors you, and I appreciate it. But it is not exactly like many true experts in repairing those very early PCs in this time and age exist anymore. At present, for nearly everyone, it's about finding your way through it, and we love watching the problem solving process, and the convoluted way that leads to an exciting FRIGGIN' WORKS finale! You're the hero of our favourite tech adventures, keep up the great work.
Love the way you apologised to the RAM! I feel old now as my first work pc was an IBM XT.
The ending result of this video blew me away. Good work Adrian!
It may sound weird, but this was a pretty exciting diagnosis process. Thanks for all the detail.
Really enjoy your troubleshooting vids, even if you go down the wrong path before finally finding the cause. 👍
Pro tip: If you don't own a thermal camera spray isopropanol on all of the chips. The ones which are hot will have the liquid evaporate the fastest.
Adrian
Don't forget to cover the window over the 27??? EPROM, as I was using them in the early days and if they were not covered we found that strong uv light would currupt the data.
We used to put a rectangular one on with programm title issue and number of EPROM.
I think the max we used was somthing like 12, but it was a number of years ago app 40.
It think it was when NASA was taking nearly all the manufacturer of special 74 series chips for the Saturn moon shot.
I think it was the 7414 schmitt-trigger ones and others.
The early IBM XT and AT models were brilliant our problem at Vauxhall (GMC) in Luton UK was sending the larg amount of data on the engine tests, back to Flint USA it was a nightmare, the systems would go down, regulary.
It would have been quicker to fly the data there.
Component level fault finding was my fortay, even used hot 🔥 fire lighters to diagnose faults on Ferranti computer boards the ceramic chips would work up to fantastic temperatures. Or freeze them.
Thanks for another informative 👍 technical video.
Dude, your modesty is admirable. However your digital hardware debug abilities are supreme. Fantastic content. Cheers
Well done Adrian. Your normal forensic analysis has worked again. Loved the video. We all go off on tangents of course.....
at 33:40 lol. hilarious! I enjoyed this a lot and I could feel your emotions here. Great video Adrian 👍🏻. Thanks a lot. Cheers, Peter
For your virtual bench, maybe try something like putting a 100 or 10 ohm in series with the ground for the scope probe? I wonder if you're getting AC transients across whatever isolation it has that's glitching into the digital side as a ground bounce. Alternatively try using a short and robust USB cable so the ground can't bounce as easily. The resistance of the GND leads to the PC basically determines how far ground could drop based on the transient.
Another option is to get a proper USB isolator that will get rid of the current path even more effectively if that's what it's doing.
Would a diode work, too?
Just found out about your channel and went down a deep dive into your videos! Love them!
I wince though, every time you handle motherboards/cards/chips without a ground strap!
Your basement temps is around 25C-26C based on some vids of yours and that would likely mean your relative humidity will be very low in a basement in Oregon.
Low RH means you are likely a walking static bomb, frying cmos/mos parts as you go!
Please check your RH and meybe think of a ground strap for one of your wrists.
Love the channel, love your work, subscribed and sharing it!
You may have gone down the wrong path, but you quickly found your way to the right path. Nice work.
You said it failed while you were using it, maybe it was a seating issue that showed up after the components warmed up and expanded. Maybe even dust in the socket? It would be fun, as others have said, to see what that rom Holtz. Maybe it is something that will work for other boards, even though EPROMs are a dime a dozen. It may in fact be un-x-orable. I always dig your videos! Keep up the great work.
When I worked at Gateway 2000 the 486DXLB motherboards were coming back bricked (no post). My coworker and I thought we would try something radical. We pulled the BIOS off of other bad boards of the same model and plugged them in the bricked ones. It fixed them every time. Oddly, service did not have a way to burn blank bios chips.
I love seeing the full experience incl. the false paths (something Usagi Electric also does a lot). I have definitely been there and given up too early. However, I can see that I probably need to get an (E)EPROM reader/programmer as that's very likely an issue I'm facing with my 1981 Nascom 2.
Falling down rabbit holes is all part of the investigative adventure & I find it rather intriguing!
You deserve to get funding from your government. Preserving the important history of computer development. Start a museum. Thanks for your efforts.
1x vote for a video attempting to fix USB issue on your modern motherboard!! Also if it is just front end ports check the connector/header, they and be annoyingly fragile. Awesome content as always! :)
Not gonna lie, even if your diagnostics were down the wrong path, seeing the process is still rewarding! :D thanks so much! :)
Excellent job troubleshooting! Had the same issue with my Apple IIe. Ended up swapping the RAM only to realize that the issue wasn't the memory but rather one of the ROMs having failed *facepalm*
Thank you for showing the problems and you actually working threw the issues. This helps teach all of us
Adrian - i love this! i am a IT tech for last 26 years and this stuff was before my time but i love it. this video makes me feel stupid coz you are so smarter than me! but its awesome. and we all make mistakes, it is cool to watch this stuff! Keep it up!
It's good to learn and breakthrough with new way of thinking, learn from mistake feel good.
Well, the sticker on the ROM is better for identification more than preventing erase, since I can't program the 2716 with the TL866II, I've switched to use EEPROM equivalent (I made adapter for 27c512 to 2716/32) , for faster erase and programming, the ultra violet thing waste lot of time, I keep some EPROM for collection but try not to use them.
Great episode (again). I wouldn't have suspected the ROM myself.
this is more like detective work.. you got trown out on wrong path, but got on track.. well done!
I worked on server boards for almost 10 years for NCR. You are working blind without logic analyzers, digital scopes, and test PROMS. And by the time the boards got to me they had passed basic tests on the Fairchild 333 bed of nails fault finder. All failures went to a specific address I could trigger on and trigger out to the digital scope to see what was going on in real time on the board.
Hi, amazing work! Could you please put these videos in playlists? I cannot afford following the videos immediatelly as they're released and I am losing context if I missed something from the subject and looking for previous videos might be a struggle. You're meaning that MB blackscreened in the past video, now I am wondering which one... I love your content and would like to keep up with your projects :)
I like learning with you Adrian. Well done, mistakes and all. Glad you don't edit that out.
When I was restoring a certain 386 motherboard I decided to populate the board with filtering capacitors where the manufacturer cheaped out on them. All the spots where the power lines, except a single spot where it was 2.5V. Turned out it was IRQ2 line and it was pulled with 10k to GND and +5V at the same time.
Maybe start with the diagnostic ROM in the first place. That helps you eliminate bigger swaths of potential faults from the get go. Good job fixing it.
I keep meaning to try figuring out what's wrong with my 5170 motherboard. It's actually got a similar issue to your 5160 board, it doesn't do anything when powered on. Now that I know that minuszerodegrees has diagnostic ROMs for these boards, I should try that out. A POST card would probably come in handy too, although I feel like it's not even getting to that point.
Regarding PALs, there are (expensive) devices to read working chips, save the contents, and write new ones. I observed a fellow with such a machine working on old classic coin op arcade games and he replaced a few in some of the videos I watched. It's rather esoteric compared to burning an EEPROM.
Always a pleasure to watch your videos! By digging for the root cause, as you said, we both learn new and interesting things. Great content!
Floating, un-terminated data/address lines can cause some interesting, intermittent, hard to diagnose failures. Many years ago I remember a particular piece of education software for the old Apple that would intermittently crash. Remember the Franklin/Apple knockoff computer? Well that same piece of education software would not run at all on the Franklin. It seems the Franklin motherboard was a bit larger than the Apple. It looked like Franklin just simply optically increased the physical size of the board but kept the IC socket sizes locations and everything else the same. I can't remember if it was a data or an address line but it ran all the way around the periphery of the motherboard. That little extra trace length on the Franklin motherboard, coupled with the floating data/address line, created a timing issue that prevented the program from running on the Franklin. Installing 2k pull-up or pull-down, (can't remember which it was) resistors on these all of these lines solved the issue for both the Apple and Franklin. This same program also had an issue with inconsistent and poor video colors caused by a missing pull-up resistor and a cap in the video modulator. (This was much worse on the Franklin than the Apple.) An added cap and a 2.2k pull-up resistor took care of this issue as well. I spent a few days adding these mods to all of the Apple II and Franklin computers used by a local, large public school system. This was a few years before they got an IT department.
If I had to guess about that empty socket, and it is completely a guess because I frankly have never even seen one of these motherboards, I’d guess part of a test harness or something used to test the board in production. The pins on the bottom side would make an ideal spot to contact on a bed of nails fixture, it it could have some kind of ROM chip in it to run test code.
Good job again Adrian, as usual it was a pleasure to watch. I sincerely learn something from it every time:)
I watch this channel for Adrian's enthusiasm when he succeeds fixing some old piece of electronics.
I really appreciate you showing the IC was backwards; makes me feel much better about the socket I put in backwards while putting the Harlequin ZX Spectrum together :-)
Hi Adrian. Really enjoyed this video. It was great. Thanks for sharing your troubleshooting process. Please keep making this type of video so we can learn with you.
For your test bench rebooting, try putting it on its own UPS to help isolate it. If that works, there are cheaper fixes than dedicating a UPS permanently, but it's an easy enough way to find out. I have a monitor that likes to power cycle every time the air conditioning switches on, so I put a power conditioner on the line, and even that only mostly (~80%) fixes the problem.
at least use a 120v to 120v isolation transformer to start....
@@GuillermoReisch That's pretty much what the "power conditioner" is in this case -- just a medium-duty 1:1 isolation transformer.
Side thought based on your comment about your lab PC having dead USB ports -- if you have dead USB ports, 'tis the season for monitor flickers and dead USB ports due to static electricity in the winter. One fix I do that works pretty reliably is to pop out the BIOS battery and make sure all the power has drained from the mobo and give it about 15 minutes, and that often resets dead USB ports. Maybe give that a shot?
Magically fantastical Infotainment - an immediate "classic" - Thank's so much! You RoCk (post editing skills were off the chart!)
It's good you have electrical engineering education and experience of some form I have been doing ee as a hobby for a while now I find myself fixing Pentium iii and IV or even earlier motherboards the newer Post lga 775 boards have their idiosyncrasies I was given one with a celeron d allegedly a PC tech said it was unrepairable well just as you do I fixed it but it was a different issue but good work mate keep the videos coming I always learning from your videos
Chips can be the most resilient piece of hardware or can fail just by touching them if you have some static on you. On that regard, I learned that the hard way with a bios on an 3d gen Intel Core motherboard about 5 years ago, and that hurt because it was my main PC. Fortunately I already had the same IC programmer as you and I ordered the same bios chip so I could repair it, but also ordered an anti static strap and I always work with it every time now.
Mistakes are the only way I truly learn. Doing things completely right can reenforce thinking that might be rightish but not completely right.
Good save. Imagine those feelings, making this repair in it's time, when these motherboards used to costs even like 1000$ in todays money.
It's fascinating to see how the diagnostic chip reported the RAM IC being reversed. I did think that was a sneaky move to prove how it can detect these changes but probably risky if it did kill it. As it turns out, they are quite resilient to this sort of thing. I'm surprised that the board didn't beep with the CGA display being absent though. I suppose this was introduced a little later on.
my 5160 (640k board with a MDA) beeps when the video memory check fails. perhaps it was indeed not present on older boards, or adrian's board is jumpered incorrectly (unlikely, because CGA works once plugged).
Adrian, if you haven't already found it, and if my info is of any use, my IBM Personal Computer XT Technical Reference (6936808), Appendix D, Logic Diagrams, page D-8, System Board (Sheet 6 of 10) shows U84 as being an LS158, a Quadruple 2-Line to 1-Line Data Selector/Multiplexor (which I had to look up). But it does show pins 1, 2, 3, 4, & 15 (GND) as being connected, and of course, it doesn't even bother showing pin 16 (VCC). Since it shows more pins connected than your board appears to have connected, I guess my logic diagram is for a different version of the System Board?
I'm much more of a novice than you are at this than you claim to be, but I learned a _lot_ from watching this video! Thanks for sharing. ;)
Adrian I'm not sure whether or not you are using the TL866 or TL866 II programmer but the software and firmware has had some pretty major updates, I just updated to version 11 and the interface has changed a bit. The logic chip database is now much larger, the programmer will now attempt to automatically identify chips (it gets it right about 90% of the time), and you can now input your own test parameters for identifying chips. Its pretty great
lol adrian the mobo on your thumbnail blends in and looks like a mobo christmas sweater
also the expression on your face makes me think of doc brown GREAT SCOTT haha
Victory is sweet! Congratulations! (and thanks for showing the whole process)
Splendid investigation!
For checking for pullups you could measure resistance against possible rails, including for unintentional pullups
This channel is gold.
We’ve all been led down the wrong path when trying to chase down a fault in a circuit board, Adrian. It can be a mix of a DOH 🤦♂️ moment followed by a “it freaking works” moment when you find the actual problem and fix it…
I had to diagnose RAM issues on a Hyundai XT clone I bought off eBay. No technical reference manuals available, so I ended up reverse engineering a good chunk of the board to point me in the right direction (based on the symptoms I was getting). Turned out to be a faulty 74x00, one of the NAND gates had gone bad and it couldn't select the active bank properly.
These are my favourite class of systems to diagnose and repair. IBM documented the truth tables for the PLAs, so one of those could be replaced if you needed. The issue, as you said, is with some of the clones like Tandy, Hyundai etc, where they'd integrated a bunch of the logic into ASICs. In which case, if you can't find a donner machine, you're probably hooped.
The osc software, the reason it faults while cycling devices on AC is because there's a sudden power spike, which can cause the power supply protection to shut off the device and keep it off till it's removed from power
There's some ways of avoiding this like an isolated powerpoint for device's your testing, a good spike suppression plug and/or power board, or your best option is a EMI/EMC/RFI inline filter
Very good troubleshooting! I totally thought it was a RAM controller chip or something just up from the RAM.
Thank you, so much, for covering the erase window on the chip.
I was gonna say (but you did cover it in the end), you filmed the death of that ROM. Watching the bus, you can see the ROM program running, then quit and halt the machine. The bus noise was just a distraction. It does make me wonder if adding a pull-up resistor to these machines would make them run more stable.
Would make them more reliable, as TTL pull up is pretty weak compared to the pull down, though better would be to add mid level terminators instead, either the 330R 470R series type, or the later integrated weak buffer CMOS type that actively held the line at 0 or 1 till a gate drove the weak buffer to the opposite state.
With LS TTL designs, open circuit buses are best left unless the design requires specific levels when the bus is not actively driven. If you pull signal lines high, you create more system noise when the lines are driven low. And mid-point termination is seldom needed. It's main purpose would normally be impedance matching, but at the edge speeds in a 74LS system with CPUs below 10MHz, it's not an issue. IBM did add terminators for the DRAM array where the distributed bus capacitance on address and control lines makes it act more like a transmission line, and that would have been normal then, just as it is today. Back in the mid-80's I did once have to put a pull-up on a memory bus on a card I had designed - the 16 floating data lines were being seen by the parity chips which were feeding back noise into the memory array and the whole thing oscillated at about 100MHz... A weak pullup (4.7k I think) fixed it.
I read the XT BIOS listings very thoroughly in the old days when I still had that kind of patience and focus, and I remember thinking the stack error would probably never happen because if there was something wrong that would cause it, it would almost 100% likely also cause either problems before that. So that was you first red herring right there.
I'm sure I mirror alot of people here when I say this, even the most experienced have moments like this. It's easy to get dragged down the wrong path, what's important out of all of that is that you did not give up and found the answer.
That right there is what separates good tech troubleshooters versus the ones that are bad :)
Fantastic, once again. I really enjoyed this. Learned so much!
Fantastic video. I always enjoy watching peoples diagnostic procedures. Can I ask - the eeprom programmer you have - is it a real mini pro? Or a TL866II clone? Are you able to post a link of where you purchased it from? Many thanks and keep up the great work :-)
Very nice video.
Very nice and lovely fixing for the first PC.
Congratulation :)
Very funny, I made the same mistake troubleshooting my Commodore PC 10. Swapped the RAMs and the 245 but still the same. In the End there was no fault, was a bad Program Version that caused Freezes. And the half height Levels we're normal. Spended two days fixing a perfectly working machine. 🤣
I see you have NCR DOS on that floppy....Is it downloadble somewhere? I still have a old NCR PC4i in the attic...